1. Field of the Invention
The present invention relates to a vision chip, and more particularly, to an optical vision chip (OVC). This work was supported by the IT R&D program of MIC/IITA. [2005-S-051-02, Photonic device integrated module for optical access network]
2. Description of the Related Art
Like human observation and recognition of objects and their identification, computer recognition of objects is a vital future industry, and also, the computer recognition of digital images will become core techniques of future robots and applicable to the automobile industry.
Research on digital image recognition has concentrated on the development of software based on a digital image processing theory rather than hardware. Such software transforms a 3 dimensional (3-D) object into a 2 dimensional (2-D) optical image, converts the 2-D optical image into an electrical digital signal, and analogizes the original 3-D object from the electrical digital signal. The image recognition method depends on such software and has a different image recognition structure from humans or animals. Thus, the image recognition method still has some initial difficulties.
The digital image processing theory uses an edge detection and recognition method to recognize images, and such methods pre-store standard models of various kinds of objects, detect an edge of an object that is to be newly recognized, compare the edge of the object with edges of the pre-stored standard models, and recognize an identity of the object.
The edge detection and recognition method used in the digital image processing theory is based on complicated mathematical calculations. In other words, the edge detection and recognition method includes a process of dividing an edge of an object into several areas, fitting edges of the several areas to a straight or curved line equation, and detecting constants of a function of the straight or curve line equation. In this method, the edge detection and recognition algorithms of objects are complicated, and also, complexity, accuracy, speediness, etc. of calculations for detecting and recognizing an edge are complicated and limited.
Semiconductor chips used in the edge detection and recognition method are referred to as vision chips, and such a vision chip has a chip structure in which a parallel operation is performed in order to minimize time required for inputting and outputting data. The vision chip also includes a photodiode (PD), a processing element (PE), and a random access memory (RAM) inside a single pixel, and the vision chip has a difficulty recognizing an image due to the limits of the digital image processing theory as described above.